What we learned about the hyperscalers from earnings season

The recent earnings season of Amazon.com Inc., Alphabet Inc. (Google), and Microsoft Corp. revealed three distinct long-term AI strategies among the hyperscalers. Amazon's AWS (Amazon Web Services) emerged as the most promising long-term winner, backed by its open, platform-first approach that offers customers a broad choice of models and silicon.

Microsoft was found to be a solid second, protected by its existing customer base, while Google was identified as the distant third due to its closed, Gemini-first strategy and a fragile business foundation.

Amazon's AWS witnessed impressive growth, growing by 39.3% at its current size, with AWS revenue reaching $46 billion and marking its sixth consecutive quarter of acceleration. The business annualized to $184 billion, nearly two-and-a-half times its size back in the fourth quarter of 2021. AWS CEO Andy Jassy projected a 40% growth rate for AWS in both 2027 and 2028, with the company aiming to reach $1 trillion in revenue within five years.

Amazon's Bedrock service played a crucial role in its strategy, offering model neutrality that allowed customers to choose from more than 100 models from various providers, including OpenAI, Anthropic, Meta, and others. The open-weight wave made this strategy even more important, as enterprises began fine-tuning smaller, cheaper models for specific tasks and reserving frontier models for the most challenging problems.

The platform that enables customers to route across all models under a single security and governance model has a significant advantage over a cloud built around a single-model company.

Microsoft's Azure grew by 48%, reaching $33.2 billion, making it the most predictable business among the three hyperscalers. The company's AI strategy focused on upselling AI to the hundreds of millions of users already using Office, Teams, and Windows products. The switch costs for customers are substantial, and Microsoft is likely to capture a significant share of enterprise AI spend regardless of the comparison of its models.

Google's strategy, centered around its Gemini model, was found to be fragile and less promising compared to its competitors. The company's closed ecosystem and heavy investment in research and development have resulted in a less attractive business proposition for enterprises.